Selective Inhibition of a Regulatory Subunit of Protein Phosphatase 1 Restores Proteostasis

See allHide authors and affiliations

Science  01 Apr 2011:
Vol. 332, Issue 6025, pp. 91-94
DOI: 10.1126/science.1201396


Many biological processes are regulated through the selective dephosphorylation of proteins. Protein serine-threonine phosphatases are assembled from catalytic subunits bound to diverse regulatory subunits that provide substrate specificity and subcellular localization. We describe a small molecule, guanabenz, that bound to a regulatory subunit of protein phosphatase 1, PPP1R15A/GADD34, selectively disrupting the stress-induced dephosphorylation of the α subunit of translation initiation factor 2 (eIF2α). Without affecting the related PPP1R15B-phosphatase complex and constitutive protein synthesis, guanabenz prolonged eIF2α phosphorylation in human stressed cells, adjusting the protein production rates to levels manageable by available chaperones. This favored protein folding and thereby rescued cells from protein misfolding stress. Thus, regulatory subunits of phosphatases are drug targets, a property used here to restore proteostasis in stressed cells.

The unfolded protein response (UPR) (13) is a component of the proteostasis network (4) that adapts folding in the endoplasmic reticulum (ER) to changing conditions. Upstream components of the UPR are the ER-resident transmembrane proteins IRE1, ATF6, and PERK, which sense folding defects to reprogram transcription and translation in a concerted manner and restore proteostasis. Activated IRE1 and ATF6 increase the transcription of genes involved in ER folding, such as those encoding the chaperones BiP and GRP94. Activated PERK attenuates global protein synthesis by phosphorylating the α subunit of translation initiation factor 2 (eIF2α) on Ser51 while promoting translation of the transcription factor ATF4. The latter controls expression of CHOP, another transcription factor, which in turn promotes expression of Ppp1r15a/Gadd34 (2, 3). PPP1R15A, an effector of a negative feedback loop that terminates UPR signaling, recruits a catalytic subunit of protein phosphatase 1 (PP1c) to dephosphorylate eIF2α, allowing protein synthesis to resume (5). UPR failure contributes to many pathological conditions (6) that might be corrected by adequate boost of this adaptive response.

Guanabenz, an α2-adrenergic receptor agonist used in the treatment of hypertension (7), shows anti-prion activity (8). We decided to assess whether guanabenz had broader activity in protecting against detrimental accumulation of misfolded proteins. Indeed, guanabenz protected against the lethal effects of expression of misfolding-prone insulinAkita (9, 10) in the ER of mouse Min6 and rat INS-1 pancreatic beta cells (Fig. 1, A and B). Guanabenz also promoted the survival of HeLa cells exposed to cytotoxic ER stress induced by tunicamycin in a dose-dependent manner (Fig. 1C), with a median effective concentration of ~0.4 μM (Fig. 1D). Unlike guanabenz, the α2-adrenergic receptor agonist clonidine and the receptor antagonist efaroxan were without effect on the survival of tunicamycin-stressed cells (fig. S1). Thus, guanabenz rescued cells from lethal ER stress by a mechanism independent of the α2-adrenergic receptor.

Fig. 1

Guanabenz protects cells from deleterious accumulation of misfolded proteins in the endoplasmic reticulum. (A) Assessment of the number of viable, blasticidin-resistant Min6 cells, after transduction with blasticidin resistance (blst)–tagged lentiviruses encoding a misfolding-prone fusion of insulinAkita with green fluorescent protein (GFP) (10), treated with the indicated concentrations of guanabenz. Values, corresponding to the cells’ ability to reduce WST-8 into formazan, were normalized to those of cells transduced with blst-tagged lentiviruses encoding cytoplasmic GFP. (B) Same as (A) with INS-1 cells. (C) Viability of HeLa cells assessed by the reduction of WST-8 [2-(2-methoxy-4-nitrophenyl)-3-(4-nitrophenyl)-5-(2,4-disulfophenyl)-2H-tetrazolium] into formazan, after treatments with tunicamycin (2.5 μg/ml; Tm) for 12 or 24 hours, with or without the indicated concentrations of guanabenz. (D) Dose-dependent protection of HeLa cells by guanabenz, from ER stress induced by 6 hours of exposure to tunicamycin. Data are means ± SD (n = 4). **P ≤ 0.001; ***P ≤ 0.0001.

We next measured the effect of guanabenz on the different branches of the UPR. Unlike tunicamycin, 50 μM guanabenz did not increase the levels of the UPR targets GRP94, BiP, ATF4, and CHOP (Fig. 2A), nor did it decrease protein synthesis (Fig. 2B). Thus, guanabenz on its own did not induce the UPR or inhibit the translation machinery. Next, we examined the effect of guanabenz in stressed cells. Attenuation of translation and increased eIF2α phosphorylation, observed 2 hours after addition of tunicamycin, were not altered by guanabenz (Fig. 2, C and D); this result shows that guanabenz did not directly antagonize tunicamycin. As previously reported (11), recovery of translation was evident 5 hours after addition of tunicamycin (Fig. 2, B and C). Guanabenz markedly delayed this translational recovery (Fig. 2C and fig. S2), sustained eIF2α phosphorylation in stressed cells, and attenuated tunicamycin-induced expression of the ER stress markers analyzed, including the pro-apoptotic protein CHOP (Fig. 2D).

Fig. 2

Guanabenz increases chaperone availability by attenuating translation recovery after ER stress. (A) Immunoblot of the indicated proteins in lysates of HeLa cells treated with guanabenz (50 μM) or tunicamycin (2.5 μg/ml; Tm) for the indicated times. (B) Autoradiogram of [35S]methionine-labeled proteins in cell lysates resolved by NuPage after a 10-min labeling pulse of HeLa cells exposed to guanabenz (50 μM) or Tm (2.5 μg/ml) for the indicated times. Lower panel is a photomicrograph of the Coomassie-stained gel. (C) Same as (B) except that cells were treated with Tm in the presence or absence of guanabenz. (D) Immunoblots of HeLa cell lysates from cells treated with tunicamycin (2.5 μg/ml) in the presence or absence of guanabenz (50 μM) for the indicated times. (E) Immunoblots of BiP recovered in SDS-resistant high–molecular weight (HMW) complexes (12) or in the total cell lysates from HeLa cells left untreated or treated with tunicamycin in the presence or absence of guanabenz. Representative results of three independent experiments are shown.

Sustained translation attenuation caused by guanabenz during ER stress is likely to increase the ratio of chaperones to substrates and favor protein folding. Indeed, the abundance of tunicamycin-induced high–molecular weight complexes, containing the ER chaperone BiP engaged with misfolded proteins (12), was markedly reduced by guanabenz (Fig. 2E). Thus, guanabenz acted like a proteostasis regulator (4) by lowering protein misfolding in stressed cells.

Two related regulatory proteins, the stress-inducible PPP1R15A/GADD34 (11) and the constitutively expressed PPP1R15B/CReP (13), recruit PP1c to form heterodimeric phosphatases that dephosphorylate eIF2α. The effects of guanabenz on stressed cells were reminiscent of those observed after ablation of Ppp1r15a (11). Analysis of the PPP1R15A-PP1c phosphatase complex revealed that PP1c was recovered in PPP1R15A immunoprecipitates from lysates of tunicamycin-treated cells, but not when cells had also been treated with guanabenz (Fig. 3A). Guanabenz similarly disrupted the complex between endogenous PP1c and overexpressed tagged PPP1R15A, whereas the related complex containing the constitutive regulatory subunit PPP1R15B remained intact (Fig. 3B). This explains why guanabenz delayed recovery of protein synthesis in stressed cells without inhibiting translation in unstressed cells not expressing PPP1R15A (Fig. 2, B and C). Guanabenz also disrupted the isolated PPP1R15A-PP1c phosphatase complex in a dose-dependent manner (Fig. 3C). Complete dissociation of all PPP1R15A-PP1c complexes in vitro occurred at higher concentrations of guanabenz than those required for cytoprotection (Figs. 1C and 3C). Thus, it appears likely that cytoprotection by guanabenz required dissociation of only a fraction of the PPP1R15A-PP1c complexes. Unlike calyculin A, which inhibits the catalytic activity of PP1c and is toxic to cells (14), guanabenz had no effect on the phosphatase activity of the isolated catalytic subunit (fig. S3) and was not toxic at the concentrations required to protect cells from protein misfolding (Fig. 1C). Thus, guanabenz did not inhibit PP1c directly.

Fig. 3

Guanabenz selectively binds PPP1R15A and disrupts the PPP1R15A-PP1c complex. (A) Immunoblots of PPP1R15A immunoprecipitates (IP) recovered from lysates of HeLa cells treated with tunicamycin for 10 hours, in the presence of guanabenz for the last 5 hours where indicated (50 μM). SN, supernatant of the immunopurification. (B) FLAG immunoprecipitates of HeLa cells overexpressing FLAG-tagged PPP1R15A or PPP1R15B, untreated or treated with 50 μM guanabenz for 6 hours, analyzed by immunoblotting. (C) FLAG-PPP1R15A immunoprecipitates from unstressed HeLa cells, washed with indicated concentrations of guanabenz and analyzed by immunoblotting. (D) Immunoblots showing that PPP1R15A but not PPP1R15B, expressed in rabbit reticulocyte lysate (input), selectively bound to biotinylated guanabenz immobilized on neutravidin beads. (E) In vitro translated PPP1R15A bound biotinylated guanabenz immobilized on neutravidin beads eluted with 1 mM guanabenz. Immunoblots show inputs, eluates, and beads before and after elution. Note that PP1c did not bind to the biotinylated guanabenz. (F) Top: Schematics of human PPP1R15A. Lower left: Same as (D) except that proteins were in vitro translated in the presence of [35S]methionine and revealed by phosphorimaging. Lower right: Coomassie-stained gel showing bacterially expressed and purified PPP1R15A230–674 fragment (input) selectively retained on neutravidin beads in the presence of biotinylated guanabenz (bound). Representative results of at least three independent experiments are shown in each panel.

To identify the target of guanabenz, we synthesized a biotinylated form (fig. S4A) with cytoprotective activity similar to that of the nonbiotinylated form (fig. S4B). When immobilized on neutravidin agarose beads, biotinylated guanabenz captured PPP1R15A but not PPP1R15B (Fig. 3, D and E), thus establishing that it selectively targeted PPP1R15A. Biotinylated guanabenz bound to a C-terminal fragment of PPP1R15A230–674 containing the PP1c binding site (Fig. 3F). A recombinant PPP1R15A230–674 fragment also bound to guanabenz, demonstrating a direct interaction between guanabenz and PPP1R15A (Fig. 3F).

The lack of both inducible and constitutive eIF2α phosphatases encoded by Ppp1r15a and Ppp1r15b is lethal in mice (15). Thus, inhibiting PPP1R15A in cells lacking Ppp1r15b should eliminate all eIF2α phosphatase activity and also be lethal. Indeed, guanabenz compromised survival of mouse embryonic fibroblasts (MEFs) lacking Ppp1r15b (Fig. 4A). However, exposure to guanabenz had no measurable effect on the viability of wild-type MEFs, or MEFs lacking a functional allele of Ppp1r15a, at concentrations that protected against protein misfolding. To further validate that guanabenz rescued cells from misfolding stress by targeting PPP1R15A, we assessed whether guanabenz could protect cells lacking PPP1R15A activity from lethal ER stress (11). Although it protected wild-type MEFs from ER stress, guanabenz had no protective effect in MEFs lacking PPP1R15A activity (Fig. 4B). Thus, all the measurable cytoprotective activity of guanabenz in ER stressed cells resulted from its inhibition of PPP1R15A.

Fig. 4

The ER stress–cytoprotective activity of guanabenz is mediated by inhibition of PPP1R15A. (A) Viability of wild-type MEFs or mutant cells (mut/mut) lacking PPP1R15A or PPP1R15B activity after exposure to the indicated concentrations of guanabenz for 48 hours. Data are means ± SD (n = 4). ***P ≤ 0.0001; n.s., not significant. (B) Viability of wild-type or Ppp1r15a mutant (mut/mut) MEFs exposed to tunicamycin (1 μg/ml; Tm) for 6 hours with the indicated concentrations of guanabenz, assessed by the ability to reduce WST-8 into formazan. The reducing activity of cells of either genotype exposed to tunicamycin without guanabenz was normalized to 1.

Our results show that specific phosphatases can be inhibited by targeting their regulatory subunits and that selective inhibition of PPP1R15A protects cells from the otherwise lethal accumulation of misfolded proteins in the ER. Guanabenz inhibits PPP1R15A and tunes translation in stressed cells to levels manageable by the available chaperones, while sparing PPP1R15B, thereby avoiding intolerable levels of eIF2α phosphorylation (15) and deleteriously low levels of protein synthesis (fig. S5). This approach to correcting proteostasis defects by inhibition of PPP1R15A could benefit many conditions characterized by the accumulation of misfolded proteins.

Supporting Online Material

Materials and Methods

Figs. S1 to S5


References and Notes

  1. We thank A. Merritt and C. Wallace for biotinylated guanabenz, J. Hastie for purified recombinant PP1c, anonymous reviewers for suggestions, and M. Goedert for advice on the manuscript. Supported by the UK Medical Research Council. D.R. is a Wellcome Trust Principal Research Fellow. A.B. is a co-inventor on patent WO/2008/041133.

Stay Connected to Science

Navigate This Article